AI drives MLCC shortage

The rapid advancement of AI is driving a shift in the delivery of key components, particularly MLCC ceramic capacitors. The adoption of 48V power lines is accelerating, as their higher voltage compared to 12V systems reduces power loss and boosts board design efficiency. At the same time, resonant circuits, such as resonant converters (LLC), are being used more and more for high-efficiency power conversion. The introduction of new 800V power systems marks the next step in AI server power architecture, with AI power systems becoming increasingly complex and sophisticated.

This is driving a shift in a key passive component, the multi layer ceramic capacitor (MLCC). The production of MLCCs is dominated by Japanese companies – Murata, TDK, Kyocera AVX and Taiyo Yuden – as well as Samsung Electro-Mechanical in Korea and Yageo in Taiwan. The AI wave is shifting from semiconductors to passive component MLCCs, says Samsung. AI servers use ten to 15 times the number of MLCCs in comparison to general-purpose servers. Not only are they driving volume, but also expanding demands into ultra high capacitance and high voltage MLCCs which require advanced technologies.

As a result major Japanese and South Korean suppliers have shifted their production capacity toward components for AI applications. This, in turn, is steadily constraining the supply flexibility of consumer MLCCs quarter by quarter. Market analysts and distributors are warning that high-capacitance MLCCs are entering a period of shortages as tier-one manufacturers divert production lines to support the rapid expansion of AI infrastructure. The situation is reminiscent of 2021 when supply of MLCCs became a critical issue for equipment makers in the Covid shortage.

The latest market research from TrendForce in Taiwan shows that the MLCC market in 2Q26 is showing a clear split between robust AI-driven demand and soft consumer demand. The Iran conflict has pushed up oil and natural gas prices, driving higher energy and transportation costs while inflation is expected to intensify and weigh on end-market demand and corporate capital spending. These effects are gradually feeding through to the electronic components supply chain. Meanwhile, rising prices for key metals such as silver, aluminium, and copper have led to price increases of 10 to 15% on average for passive components.

The tight supply of key components has prompted OEMs such as Dell and HP to adopt strategic inventory-building measures. Some low- to mid-tier notebook orders originally scheduled for 3Q have been pulled forward into 2Q production and shipment to lock in lower pricing expectations, stimulate demand, and boost revenue. In response to shrinking production capacity and strict inventory controls for consumer MLCCs, agents based in Taiwan and Mainland China have begun pre-emptively stocking X5R standard products (capacitance values between 1000pF and 10µF).

MLCC shortage

In April 2026, Taiyo Yuden took the lead by raising prices for low-capacitance consumer and automotive MLCCs by 6% to 13%. By early May, some equipment makers had already completed their 3Q26 price negotiations with MLCC suppliers, with process recovering. As the majority of ODMs begin a new round of price negotiations in late May, a key question is whether current market conditions drive a rebound in MLCC prices for automotive and consumer applications. Mordor Intelligence values the Low Voltage MLCC market at US$6.17 billion in 2025 and estimated to grow from $18.97 billion in 2026 to $42.18 billion by 2031, a significant CAGR of 17.33% for a passive component.

Demand in 5G handsets, AI servers and electrified vehicles require ultra-compact, low-loss capacitors that can be mounted in ever-tighter spaces while delivering higher capacitance per volume. While Asia-Pacific retains its manufacturing and consumption lead, near-shoring and semiconductor stimulus programs are pushing new capacity closer to end customers in North America. The leading manufacturers are expanding production of layers less than 0.3 µm thick to further reduce the size of the MLCCs without sacrificing reliability.

This demand has already been reflected in recent shipments and revenue performance of contract equipment makers such as Quanta, Wistron, and Compal. However, since OEMs have not revised their full-year shipment forecasts upward, risks are building for a weaker-than-usual peak season in the second half 26, says TrendForce, along with potential order corrections. Strong demand from AI servers has prompted Japanese and Korean manufacturers to reallocate capacity from consumer-grade products to high-end MLCCs. The industry book-to-bill (BB) ratio improved from 0.89 in March to 0.92 in April, while the leading suppliers have maintained ratios consistently above 1, indicating expansion.

Taiyo Yuden has already raised prices for low- to mid-capacitance consumer MLCCs and select automotive products by approximately 6 to 13% for distributors in China. Meanwhile, Yageo and Walsin Technology Corporation (WTC) are negotiating price adjustments on a case-by-case basis for certain loss-making products, without announcing broad price hikes. Industry leaders Murata and Samsung Electro-Mechanics have yet to make formal announcements, but overall pricing sentiment is shifting from cautious observation to tentative upward adjustments. TrendForce expects that AI server projects that are ramping up at the end of the year will tighten supply of high-end MLCCs and drive up prices.

AI

High capacitance MLCCs are increasingly in demand to support both GPUs and CPUs that consume thousands of amperes of current at a low voltage of 0.8V. The total capacitance of the MLCC has to increase to ensure a stable power supply for GPU power Dynamics. A Grace Blackwell GB200 server board from Nvidia requires approximately 6,500 MLCCs, while the coming Rubin architecture will increase this to roughly 12,000 devices per board. Simultaneously, cloud service providers (CSPs) such as Microsoft, AWS, Google, and Meta are steadily increasing orders for in-house ASIC chips and CoWoS advanced packaging, driving up long-term demand for high-end MLCCs.

MLCCs placed near high-performance computing board GPUs and CPUs play a decoupling role to mitigate rapid current changes. As the performance of chips increases, the mounting area is reduced, whilst the required capacitance grows. This requires capacitances of over 47µF within 0402 inch size or 10µF in 0603 inch size. The capacitance of MLCCs which undergo surface mount process near GPU balls is also increasing. Embedded MLCCs and landside MLCCs that are mounted within the semiconductor package or right underneath reduce loop inductance significantly while increasing capacitance density.

Another factor is power efficiency. While 48V systems require 100V MLCCs, a 120kW power delivery system for AC to 800V conversion will need larger 1kV–2kV MLCCs. In response to GPU load current dynamics, large core power supply currents are necessary. VPD (Vertical Power Delivery) technology is a form of power module that not only shortens the power path as much as possible, but also increases power density.

Murata

Last year Murata Manufacturing started the world’s first mass production of the 0402 size MLCC that measures 1.0 × 0.5 mm with a capacitance of 47µF for AI systems. Equipment in AI servers and data centre requires increasingly high component density, so optimized component placement within limited PCB areas is paramount. As a result, there is increasing demand for capacitors that offer both miniaturization and higher capacitance, along with high reliability under high-temperature conditions caused by heat generated from PCBs and integrated circuits (ICs).

Murata used its proprietary technologies in ceramic dielectric layers and internal electrode miniaturization for the production of the 47μF MLCC. Compared to Murata’s conventional 0603-inch size product with the same capacitance, the capacitor reduces mounting area by approximately 60%. It also delivers over twice the capacitance of Murata’s previous 22μF product in the same 0402-inch size. The MLCC is available in two variants – the X5R (EIA) GRM158R60E476ME01 with an operating temperature range of -55 to +85°C, and the X6S (EIA) GRM158C80E476ME01 at -55 to +105°C.

Kyocera

Moving AI into smartphones is also driving demand for MLCCs. At the end of 2025 Kyocera started production of a 47µF capacitor in the 0402 form factor measuring 1.0mm x 0.5mm. The 0402 is the most widely adopted size for smartphones and wearable devices and Kyocera has reduced the thickness of the dielectric and internal electrodes, expanding the capacitance per unit by approximately 2.1 times compared to the previous 22μF capacitor in the same size, reducing the number of components. Increased thermal resistance up to +105°C ensures high reliability in harsh temperature environments such as AI servers.

Taiyo Yuden

Taiyo Yuden in Japan has the third largest market share of MLCCs, and its latest embeddable MLCC achieves a capacitance of 22μF in a 1005 size (1.0 x 0.5 mm). This is designed for